On Wed, Feb 14, 2018 at 12:22:02PM -0500, David Miller wrote:
> From: Andrew Lunn <and...@lunn.ch>
> Date: Wed, 14 Feb 2018 14:40:40 +0100
>
> > linux/drivers$ grep -ir aspm * | grep MODULE_
> > gpu/drm/amd/amdgpu/amdgpu_drv.c:MODULE_PARM_DESC(aspm, "ASPM support (1 =
> > enable, 0 = disable, -1 = auto)");
> > gpu/drm/radeon/radeon_drv.c:MODULE_PARM_DESC(aspm, "ASPM support (1 =
> > enable, 0 = disable, -1 = auto)");
> > infiniband/hw/hfi1/pcie.c:MODULE_PARM_DESC(aspm, "PCIe ASPM: 0: disable, 1:
> > enable, 2: dynamic");
> > net/wireless/realtek/rtlwifi/rtl8192ee/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8188ee/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8821ae/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8192se/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8192ce/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8723be/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8192de/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > net/wireless/realtek/rtlwifi/rtl8723ae/sw.c:MODULE_PARM_DESC(aspm, "Set to
> > 1 to enable ASPM (default 1)\n");
> > staging/rts5208/rtsx.c:MODULE_PARM_DESC(aspm_l0s_l1_en, "enable device
> > aspm");
> > staging/rtlwifi/rtl8822be/sw.c:MODULE_PARM_DESC(aspm, "Set to 1 to enable
> > ASPM (default 1)\n");
> >
> > This patch seems to have the exact opposite of everybody else already
> > does.
> >
> > Maybe you can follow the AMD example, and default to -1, since you are
> > proposing to mostly have it enabled, but disabled in one case?
>
> This is just another good reminder of why module parameters are a
> terrible user experience for just about anything device configuration
> related.
Agreed.
I took a look at see if there was some other way to do this. Maybe the
pcie core had a callback or something. Nope.
> For users, this leads to misery.
At least in this case, the misery is mostly limited to realtek
users. Nearly everybody else seems to get it right and not need such a
hack.
Andrew
